Every one minute seems... excessive. I know this is social media but still, what would be bad about reducing that to like once per hour, to reduce the load across the entire Fediverse?
True, but if we agree that every second is too frequent, but that every day is too sparse (maybe?), then the question becomes one of optimizing the choice of what frequency would be best for whatever outcome.
Like if it was every 5 minutes, that would cut down on the amount of traffic 5-fold, and still give 12 checks every hour. If we do not expect checks to fail, that may still be 6-12-fold more often than necessary. Or a check every 10 minutes is still "many checks per hour", and yet does not blow up the traffic stats 10-fold more than necessary to get a "desirable" outcome.
There are also additional costs to consider when doing a check every minute: that is 1440 checks every 24 hours! Data must be stored, CPU time spent, waste heat is generated (it is not exactly exasperating climate change but it is not not doing that either!), etc. Whereas if a check every 10 minutes is used, then the same number of checks would span a 10-day timeframe. Depending on whatever else the machine is doing (gaming? other server checks?), it reduces the load 10-fold and something that much may even change things on a qualitative rather than merely quantitative basis.
Well, it was a thought for consideration at any rate.
It doesn’t have the same problem as lemmy.fhmy.ml, but especially because of current unknowns around .ml domains, it would benefit greatly from independent monitoring.
Please add lemmy.fmhy.ml to that list - it’s been down for over a day due to the .ml issues but have been looking for something like this to see if its back.
Add comment